A two-layer network approach to Information Disorder
Note: You may notice that this post is dated older than when this website was created. That is because, this is a cross post. I had originally posted this in my old blog here. I am cross-posting it here in the spirit of collecting everything I have written in one place.
Information disorder, popularly referred to by its problematic synonym “fake news”, is turning out to be a massive problem. Here I try to explain a framework with which I have been contextualizing the various aspects of the problem. I believe there are two layers of networks involved - one of information and another of people - and these two networks have interactions between and within them. I am going to use some jargon on Networks dynamics which I define in my earlier blog here. So if the jargon seems foreign, kindly read through that post first.
The two layers, separately
As aforementioned, I think there are two layers of networks involved in the Information Disorder problem - The Information layer and The People layer.
- The Information layer - In this layer, the nodes are different pieces of information while the edges relate two pieces of information. The different pieces of information on a social network could be posts (aka tweets or threads) and comments (aka retweets). Two pieces of information can be related if their topics are related. Obviously, if one piece of information lead to the other piece being created (like a tweet leading to a retweet) they will be connected since they would be on the same topic.
- The People layer - As the name suggests, this layer is made of people. This is the network that most of us think about when thinking of a social network - nodes are people and edges exist between them if they are acquainted with each other in some abstract way (‘Friends’ on Facebook, ‘Followers’ on Twitter etc.).
The two layers, together
Now that we have established each layer’s nodes and the interactions within them, lets think about the interactions between the layers. A user in the People layer and a piece of information in the Information layer are connected if the user engages (creates, shares, likes etc.) with the piece of information. This interaction now creates a new type of edge within both the layers. In addition to the edges that are native to the Information layer, due to the interactions between the layers, a new type of edge is formed if the one user in the People layer engages with two different pieces of information. Similarly, in addition to the native edges of the People layer, the inter-layer interactions creates a new type of edge in that layer, if a piece of information is engaged with by two different users.
The complicatedness of the problem should already be clear given the structure of the interactions between and within these two layers I have put down so far. (See my crude depiction of the explanation so far if you are unclear). But the fun doesn’t stop there! One needs to think about the dynamics involved now.
In the People layer, dynamics of the network plays out, if a new person joins the network or existing people leave it or people make new friends or unfriend existing ones. In the Information layer, it plays out, if a new piece of information is created or a piece of information is retracted or one piece of information suddenly gets new context relating it to other pieces of information. So dynamics of the networks is straight forward in that it happens independent of the other layer. But the astute reader would have noticed a small nuance in the dynamics of the Information layer - how exactly can a piece of information be retracted? I will come to this later.
This is not the case for dynamics on the networks. In the Information layer, popularity diffuses through the network due to the non native edges created between two pieces of information even if the two pieces of news are seemingly unrelated. In the People layer, information gets debated though the non native edges created between two people who are actually not acquainted with each other at all. This dynamics is a bit more complicated than the dynamics of the networks here, since it depends on the edges created by interlayer interactions. But, and this you will recognize is a refrain in Information Disorder literature… it gets more complicated.
Now consider a user who has created a piece of content or is particularly passionate about one. To increase its popularity they could create bots in the People layer that are subscribed to their favourite piece of content in addition to other pieces of content that are more popular than their own and hence create artificial non native edges in the information layer, hence spreading its popularity. Here dynamics of the network in one layer ended up creating dynamics on the other layer’s network. This is a case of hybrid dynamics. There are several other examples of hybrid dynamics playing out. I leave it to you as a thought exercise to imagine them.
Contextualizing the problem using the two layer framework
Lets now try to use these two layers to contextualize the solutions that are typically proposed to solve for Information Disorder. As I alluded to in the first blog, critical thinking/media literacy efforts focus on affecting dynamics on the People layer. When platforms put labels on posts (like ‘This claim is disputed’), they are trying to acknowledge the fact that a given piece of information may not be trending on its own merits but rather because of the artificial non native edges or the interlayer interactions and hence they affect the dynamics on the Information layer. Proposals to split up big social media firms try to address the dynamics of the People layer, with the (in my opinion, unrealistic) hope that competition created will make the size of the People layer on a given platform small. Calls to ban certain pieces of information address dynamics of the Information layer.
One can even contextualize other frameworks proposed to think of the problem within this two layer framework. To me, the First Draft’s intent-based frame for Information Disorder  seems to be a focus on the people layer. ‘Do people share information amidst themselves to harm each other or lie to each other or neither?’ seems to be its premise. On the other end of the spectrum, that use the Information Ecosystem as a framework seems to be more of a focus on the Information layer.
Is it useful, though?
So this two layer model puts into context many of the solutions and also other frameworks. But just because a model captures all the aspects of a problem it doesn’t mean it needs to be useful to solve it! (See the wonderful Jorge Luis Borges story ) Is this too complex of a model to consider for the problem or is it too simple or is it the right amount of complex? I think the answer to this question can be found only if the framework is actually used to research the problem and see if it can help in solutioning. I have seen some literature that effectively use multi layered models for other problems . I am hoping it will be useful here too… finger crossed.
 C. Wardle and H. Derakhshan, “INFORMATION DISORDER : Toward an interdisciplinary framework for research and policy making Information Disorder Toward an interdisciplinary framework for research and policymaking,” 2017.
 Borges, Jorge Luis. “On the exactitude of science. Collected Fictions.” Translated by Andrew Hurley. New York: Penguin (1998): 325.
 Poledna, Sebastian, José Luis Molina-Borboa, Serafín Martínez-Jaramillo, Marco Van Der Leij, and Stefan Thurner. “The multi-layer network nature of systemic risk and its implications for the costs of financial crises.” Journal of Financial Stability 20 (2015): 70-81.